A comparison of psychophysiological detection of deception accuracy rates obtained using the counterintelligence scope polygraph and the test for espionage and sabotage question formats

This 1995 Department of Defense study compared the accuracy of the Test for Espionage and Sabotage (TES) format against traditional Counterintelligence Scope Polygraph (CSP) formats using directed-lie and probable-lie comparison questions. The research addressed the poor detection rates of the CSP format and evaluated whether the standardized TES methodology could improve counterintelligence screening accuracy.

The study used a between-subjects design with 277 examinees recruited from communities near Ft. McClellan, including 90 programmed guilty participants who enacted mock espionage scenarios, with examinations conducted by 18 certified government examiners.

This study addressed a critical credibility crisis in counterintelligence screening where traditional CSP formats showed unacceptably low guilty detection rates (43.2% in prior studies). The research compared three polygraph formats using a controlled laboratory design with realistic mock espionage scenarios. Results demonstrated that the TES format, featuring standardized procedures, directed-lie controls, and reduced question complexity, significantly outperformed traditional CSP formats in detecting deception. The TES achieved 83.3% guilty detection accuracy versus approximately 55-58% for CSP formats, representing a substantial improvement for national security screening applications.

These findings provided empirical support for adopting the TES format in operational counterintelligence screening, demonstrating that standardized procedures and directed-lie controls can substantially improve detection accuracy. The research informed policy decisions regarding polygraph methodology in high-stakes national security contexts.

Background & Context

In the early 1990s, counterintelligence screening faced a critical credibility crisis. Prior analog studies suggested that when the Counterintelligence Scope Polygraph (CSP) format was utilized, 94.9% of programmed innocent examinees were correctly identified, but only 43.2% of programmed guilty examinees were correctly identified—an unacceptably low detection rate given the stakes of national security screening. This research emerged during a period of intense scrutiny of polygraph practices, particularly after several high-profile espionage cases revealed security vulnerabilities.

The Department of Defense Polygraph Institute developed the Test for Espionage and Sabotage (TES) as a comprehensive redesign of security screening methodology. The TES format differs in many ways from the CSP and is generally more standardized and less intrusive than older methods, and makes use of directed-lie comparison questions. This 1995 study represented the first systematic evaluation comparing TES against both probable-lie and directed-lie versions of the traditional CSP format, addressing fundamental questions about whether standardization, directed-lie controls, and modified question structure could substantially improve detection accuracy.

Research Design & Methodology

The 277 examinees included in the analyses were recruited from communities surrounding Ft. McClellan, AL, with 90 examinees programmed guilty by enacting one of four possible mock espionage scenarios, and examinations conducted by 18 certified government examiners. Twenty-nine additional examinees were excluded because eight PG examinees confessed their guilt prior to examination, six were not medically suitable, four examinations were incomplete, three were DoDPI employees, and eight FPWJ examinees were excluded. The study employed a between-subjects design where each participant received only one test format.

Three distinct polygraph formats were compared: CSP with probable-lie control questions (CSP-PLC), CSP with directed-lie control questions (CSP-DLC), and the new TES format. Examiners used standard field polygraph instruments manufactured by either Lafayette or Stoelting, recording standard respiratory, electrodermal, and cardiovascular responses. Tests were scored using the 7-point scoring criteria taught at DoDPI, in which the relative strength of physiological responses to a relevant question is compared to responses to a control question.

Key methodological features of the TES format included:

• Reduced number of issues tested in each question series
• Maximum of three question repetitions for calculating scores
• Elimination of between-test stimulation procedures
• Standardized, unalterable question sequence
• Brief, standardized pretest following logical information presentation
• Exclusive use of directed-lie comparison questions

The PG examinees enacted one of four mock scenarios, each representative of one of the four relevant questions, including an "espionage" scenario requiring one examinee to steal a classified document and give it to a second examinee. This design created realistic psychological pressure while maintaining experimental control.

Results & Key Findings

When conclusive decisions were made (inconclusive decisions excluded), the decisions of examiners who administered the TES format were significantly more accurate at 83.3% in identifying programmed guilty examinees than were decisions of examiners who administered either the CSP-PLC (55.6%) or the CSP-DLC (58.6%) format. This represented a 27.7 percentage point improvement over CSP-PLC and a 24.7 percentage point improvement over CSP-DLC for detecting deception.

The key accuracy findings were:

• TES guilty detection: 83.3% (significantly superior to both CSP formats)
• CSP-PLC guilty detection: 55.6% (baseline traditional format)
• CSP-DLC guilty detection: 58.6% (no significant improvement over PLC)
• Innocent identification across all formats: 88.9% to 95.3% with no significant differences

The accuracies of decisions obtained using the three formats to identify programmed innocent examinees were 88.9%, 95.3%, and 95.2% for TES, CSP-PLC, and CSP-DLC formats, respectively. Critically, there were no significant differences among the accuracies of examiners' decisions identifying the programmed innocent examinees, demonstrating that TES achieved superior guilty detection without sacrificing innocent protection.

Results were supported by blind scoring accuracies, with blind raters' decisions using the TES format similar to original examiners' decisions, and blind examiners significantly more accurate (81.0%) in correctly identifying PG examinees when data were collected with TES format.

Discussion & Significance

This study provided empirical validation for a fundamental redesign of counterintelligence screening that would reshape Department of Defense polygraph practice. According to the DoDPI annual report to Congress for fiscal year 1994, by that time, "All DoD agencies [had] switched from the CSP to TES for security screening"—a policy shift that this research directly supported. The magnitude of improvement in guilty detection (nearly 50% increase in accuracy rate) was unprecedented in screening polygraph research.

The findings suggested that multiple simultaneous procedural changes—standardization, directed-lie controls, reduced issue scope, and modified question structure—produced synergistic benefits. The number of programmed innocent examinees correctly identified using CSP-DLC was not significantly different than when either CSP-PLC or TES was administered, indicating that differences in accuracy rates detecting PI or PG examinees were not attributable to differences between PLCs and DLCs alone. This demonstrated that the TES advantage derived from its comprehensive systematic approach rather than any single element.

The research addressed longstanding criticisms of security screening polygraph. Where earlier CSP studies showed unacceptably low guilty detection rates, TES achieved accuracy levels more comparable to specific-issue criminal testing while maintaining the multi-issue coverage required for counterintelligence screening. This represented a critical advancement in balancing operational requirements with scientific rigor.

Limitations & Considerations

A criticism concerned the wording of relevant questions, as examinees were asked if they had committed espionage or sabotage "against the United States," and many experienced examiners believed that because PG examinees had participated in a "mock" crime and had not committed any act "against the United States," the question wording might have reduced the psychological significance of the acts that PG examinees did commit, potentially contributing to low accuracy rates for identifying PG examinees. This fundamental issue of ecological validity affects all analog polygraph research but may have been particularly pronounced in this security screening context.

Additional methodological considerations include:

• Laboratory simulation may not replicate the psychological pressure of actual espionage or screening stakes
• Paid volunteers differ from actual security clearance applicants in motivation and baseline anxiety
• Examiners aware of the experimental context may have altered their behavior
• The study could not assess long-term deterrent effects or real-world base rates

Further testing was required to answer questions raised about whether the caveat "during this project" affected accuracy of decisions identifying innocent and PG examinees, whether the question caveat impacted generalizability to field situations, whether the reduced number of relevant issues contributed to increased accuracy, and whether results would generalize when different issues and questions were utilized.

Practical Applications

This research directly informed the complete overhaul of Department of Defense counterintelligence screening practices, affecting tens of thousands of annual examinations. The TES format became the standardized approach across DoD agencies and later influenced Department of Energy and other federal screening programs. For polygraph examiners, the study validated the importance of standardized procedures, structured pretest interviews, and careful attention to comparison question construction—principles that extended beyond security screening to other testing contexts.

For individuals facing counterintelligence screening, the findings offered both reassurance and caution. The high innocent identification rates (88-95%) across all formats suggested that truthful examinees had strong odds of passing. However, the substantially improved guilty detection rate meant that the TES posed greater risk to those attempting deception. The research underscored that procedural details—question structure, standardization, comparison question type—materially affect outcomes, highlighting the importance of examiner training and protocol adherence in ensuring both security effectiveness and individual fairness.